The project "High-Fidelity Compressible Flow Simulation" is the combined effort of six research projects carried out at the Division of Fluid Dynamics/Department of Mechanics and Maritime Sciences/Chalmers University of Technology with focus on detailed simulations of high-speed engine-exhaust flows and turbomachinery flows. In all of these six projects, high-fidelity Computational Fluid Dynamics (CFD) methods are used for the simulation of flows in engineering applications with the aim to gain a better understanding of the underlying physical principles and to push the limits for industrial use of CFD. All simulations are done using the Chalmers-developed CFD code G3D::Flow, which has been used extensively on SNIC resources in the past and shown very good scaling properties.
The flow simulation methodologies applied in the project are based on (Unsteady) Reynolds-Averaged Navier-Stokes ((U)RANS), Large Eddy Simulation (LES) and Delayed Detached Eddy Simulation (DDES). Out of these methods (U)RANS implies the highest degree of modeling and is consequently the least expensive approach. The most expensive method is LES, where only the smallest turbulence scales are modeled. DDES is a hybrid method combining the ability of the more expensive LES method with the less expensive URANS method. In this methodology, URANS is used close to the walls where proper LES would require really high resolution and LES is used in the off-wall regions. In order to accelerate URANS simulations of turbo machinery flows, a time-spectral approach known as the Harmonic Balance method has been implemented in G3D::Flow. Despite the use of Harmonic Balance, large-scale turbomachinery simulations require in the order of 65 000 core hours for one single design and operating condition. DDES and LES, where Harmonic Balance may not be applied, typically require substantially larger resources.
The computational effort needed varies between the different research projects but all activities are in the large-scale simulation range, which stems from the use of computational grids with sizes in the order of 100 million cells or, depending on method, problems with roughly 700-2500 million degrees of freedom. Access to large-scale computational resources is needed in order to be able to realize the investigations signed up for in ongoing and planned research activities. Based on the need and the current level of usage, allocation of a SNIC-Medium project is justified.